genomics and drug development - ucl · director, ucl institute of cardiovascular science professor...
TRANSCRIPT
Aroon HingoraniDirector, UCL Institute of Cardiovascular ScienceProfessor of Genetic EpidemiologyUniversity College [email protected]
UCL Hospitals National Institute of Health ResearchBiomedical Research Centre
Genomics and drug development
Hingorani A D et al. BMJ 2010;341:bmj.c5945
©2010 by British Medical Journal Publishing Group
Human Genetic Variation
Genetic spectrum of human disease
Monogenic disorder
Mutation
Polymorphism
Disease
X Y
X Y Other genes
Environment
DiseaseHealth
SinglenucleotidePolymorphism(SNP)
XY
Other genes
Environment
DiseaseHealth
Polygenic disorder
Genomics and the Whitehall II study
• Creation of a DNA biorepository
• Targeted genotyping
• Cardiochip - 50,000 variants
• Metabochip – 200,000 variants imputed to 2M
• DrugDev Array – 480,000 variants imputed to ?
• The UCLEB Consortium
University College-London School-Edinburgh-Bristol (UCLEB) Consortium
Phenoytpes: wide coverage of organs/systems allows efficacy and safety profiling for most commondisorders
> 52 blood markers in up to 27,000 samples
~ 216 NMR metabolomic traits in >11,000 samples and leveraged funding for additional 15,000 samples
Organ/System Phenotype Associated disease outcome
Brain: Cognitive function Alzheimer’s disease
Heart: ECG traits AF, sudden death
Blood: Ultra-dense lipids Type-2 diabetes
Blood vessel: Carotid atherosclerosis Atherosclerotic vascular disease
Lung: FEV1, FVC COPD
Kidney: eGFR End-stage renal disease
Liver: AST, ALA, GGT Fatty liver and chirrosis
Bone: Bone mineral density Osteoporosis
Clinical events
>5000 CVD events including CHD and stroke.
Additional CV events: angina, heart failure, and DVT/PE
>4000 cancer events
>2000 Type-2 diabetes cases
~2000 COPD cases
Genomics and Drug Development -Overview
• Process of drug development and the potentialbenefits of genomic support for drug targetselection and validation
• The Druggable Genome
• Design of a new genotyping array to support drugdevelopment
Pre-clinical development Clinical development
Late-stage failure
Programme attritionCost
Drug development process
After:Kola & Landis, Nature Reviews Drug Discovery 2004; 3, 711-716Arrowsmith, Nature Reviews Drug Discovery 2011; 10, 328-329; andNature Reviews Drug Discovery 2013; 12, 569PaulS et al. Nature Reviews Drug Discovery 2010; 9, 203-214
Pre-clinical development Clinical development
Late-stage failure
Poor predictive accuracyof preclinical studies
Definitive target validationexperiment (the phase III RCT)
is the final step
Drug development process
Randomised controlled trial (RCT; Phase III)
Patients
Randomisation
Intervention Placebo
Target affected Target unaffected
Outcome Outcome
The RCT is the pivotal drug target validation experiment
Design feature Attribute
In humans Avoids limitations ofexperiments in cells,isolated organs and animalmodels
Randomised experimentalintervention
Overcomes confoundingand reverse causationinherent in humanobservational studies
Pre-specified efficacy andsafety outcomes, carefulsample size determination
Low risk of false positivefindings
RCT (Phase III)
Patients
Randomisation
Genetic studies as Nature’s randomised trials
Mendelian randomisation trial
Population
Random allocation of alleles
Target genotype aa Target genotype AA
Target activityunchanged
Outcome Outcome
Target expression or activitymodified
Variants of a gene encoding a drug target, allocated at random at conception,that affectits expression or function,
can be used as a tool to infer the outcome of modifying the same target pharmacologically
Hingorani A, Humphries S. Lancet 2005; 1906–1908
Intervention Placebo
Target affected Target unaffected
Outcome Outcome
Target protein
Intendedoutcome
On-target effect
Encodinggene
Compound
Relationship between gene, target and compound
Target protein
Intendedoutcome
On-target effect
HMGCR
Relationship between gene, target and compound
Statin
HMG-coA reductase
RCT (Phase III)
Sample
Randomisation
HMG-CoA red inhibitor Placebo
LDL-C reduced LDL-C unchanged
CV eventrate lower
CV eventrate higher
Protein target: HMGCR
Off target
Mendelian randomisation Trial
Population
Random allocation of alleles
HMGCR aa Genotype AA
LDL-C unchanged
CV eventrate lower
CV eventrate higher
LDL-C reduced
Protein target: HMGCR
HMGCR variant (rs12916)
LDL-C reduced by 0.07 mmol/L
CHD risk reduction 6%.
Ference et al. J Am Coll Cardiol 2012; 60(25):2631-9
HMGCR inhibitors (statins)
LDL-C reduced 1 mmol/L
CHD risk reduction 25%
CTT Lancet 2010, 376, 1670–1681
HMGCR variants, statins, LDL-C and coronary events
Common genetic variants and small phenotypic effectsize
0.06 mmol/Lper allele
Courtesy Daniel Swerdlow
Genomic support for drug target selectionand validation: selected examples
Pre-clinical development
MR trials: Example 1 –PLA2G2A, sPLA2, varespladib and
CVD events
Clinical trials
Progression of a new therapeutic at acritical decision point
sPLA2, Varespladib and VascularEvents: Phase-III trial (JACC 2013)
Holmes MV et al. JACC 2013 Nov 19;62(21):1966-76
Summary findings pre Phase-III trial: sPLA2-IIA concentration and activity is associatedwith incident and recurrent major vascular events
Drug-target and therapeutic: a small molecule sPLA2 inhibitor (Varespladib; Anthera)reduced sPLA2 mass by ~90%
VISTA-16: Randomised 5145 patients with acute coronary syndrome to varespladib500mg daily or placebo. Outcome was assessed at 16 weeks. The trial was stopped atprespecified interim analysis for futility or possible harm.
Summary results (http://www.anthera.com/VISTA-16.pdf)
HR for primary outcome (CVD death, non-fatal MI, stroke): 1.24, p=0.155HR for stroke 1.43, p=0.025
HR for non-fatal MI 1.68, p=0.009
Varespladib and Cardiovascular Events in Patients With an Acute Coronary Syndrome
JAMA. 2014;311(3):252-262. doi:10.1001/jama.2013.282836
General Population: Incident eventsMajor vascular events
Nonfatal MINonfatal StrokeFatal MI/Stroke
General Population: Prevalent eventsMajor vascular events
MIStroke
Acute Coronary Syndrome: Recurrent eventsMajor vascular events
Nonfatal MINonfatal StrokeFatal MI/Stroke †
OutcomeSetting,
13 (8021/56359)13 (4208/51016)11 (2304/46790)12 (1509/48118)
12 (7513/55523)12 (6411/54884)8 (1102/37280)
9 (2520/15768)8 (1158/14152)6 (223/12283)9 (1139/15724)
(events/participants)Studies
1.02 (0.98, 1.06)1.04 (0.98, 1.10)1.00 (0.93, 1.07)1.01 (0.93, 1.10)
0.99 (0.95, 1.03)0.98 (0.93, 1.03)1.03 (0.93, 1.15)
0.96 (0.90, 1.03)0.99 (0.89, 1.09)0.85 (0.69, 1.06)0.96 (0.87, 1.06)
allele) (95% CI)Odds ratio (per
26(0,51)22(0,59)19(0,59)41(0,70)
38(0,63)52(7,75)0(0,67)
0(0,45)28(0,67)0(0,74)0(0,64)
(95%CI)I2,%
1.02 (0.98, 1.06)1.04 (0.98, 1.10)1.00 (0.93, 1.07)1.01 (0.93, 1.10)
0.99 (0.95, 1.03)0.98 (0.93, 1.03)1.03 (0.93, 1.15)
0.96 (0.90, 1.03)0.99 (0.89, 1.09)0.85 (0.69, 1.06)0.96 (0.87, 1.06)
allele) (95% CI)Odds ratio (per
26(0,51)22(0,59)19(0,59)41(0,70)
38(0,63)52(7,75)0(0,67)
0(0,45)28(0,67)0(0,74)0(0,64)
(95%CI)I2,%
Lower Higher
1.5 1 2
Odds ratio
Association between PLA2G2A rs11573156and CVD outcomes (per C allele)
Holmes MV et al. J. Am Coll Cardiol 2013 Nov 19;62(21):1966-76
MR Trials: Example 1 - PLA2G2A rs11573156 allele and CVD outcomes
Target protein
Compound
Intendedoutcome
On-target effectOff-target effect
Unintendedoutcome
Unintendedoutcome
Other protein
Encodinggene
MR Trials – distinguishing on from off-target effects
Target protein
Compound
Intendedoutcome
On-target effectOff-target effect
Unintendedoutcome
Unintendedoutcome
Other protein
Encodinggene
Target profile
Compound profile
MR Trials – distinguishing on from off-target effects
Drug Randomisation
TRGHDL
No change in lipids
Sample
Torcetrapib Control
CETP-inhibition No-CETP inhibition
LDL
Change in lipid traits
BP(Off-target)?
Trait RCTs(individuals)
Torcetrapib/atorvastatinvs
Atorvastatin alone
Mean difference (95%CI)
HDL-C (mmol/L) 17911 0.78 (0.68, 0.87)
Systolic BP(mmHg)
17911 4.471(4.09, 4.84)
Hazard ratio (95%CI)
CVD events 15067 1.25 (1.09,1.44)
BP(On-target)?
Sofat R et al. Circulation 2010 Jan 5;121(1):52-62
MR trials:Example 2 – CETP, torcetrapib,HDL-C and BP
CETP gene variants, lipids and BP
2010 Jan 5;121(1):52-62
CETP Lipids and apolipoproteins
Blood pressure
Sofat R et al. Circulation 2010 Jan 5;121(1):52-62
The BP raising effect oftorcetrapib is off-target
CETP gene variants, lipids and BP
2010 Jan 5;121(1):52-62
CETP Lipids and apolipoproteins
Blood pressure
Sofat R et al. Circulation 2010 Jan 5;121(1):52-62
The BP raising effect oftorcetrapib is off-target
CETP gene variants, lipids and BP
2010 Jan 5;121(1):52-62
CETP Lipids and apolipoproteins
Blood pressure
Sofat R et al. Circulation 2010 Jan 5;121(1):52-62
The BP raising effect oftorcetrapib is off-target
MR trials: Example 3 – potential repurposing
Clinical trialsPre-clinical development
RepurposingIL6R blockade (tocilizumab)
and CHD(Lancet 2012)
Inflammation strongly linked to CHD butno currently validated therapeutic target
Clinical trialsPre-clinical development
Rheumatoid arthritis
CHD
Drug interventionPatients with rheumatoid arthritis
Randomisation (Tocilizumab)
IL6R- blocker (MAB) Placebo
Reduced IL6 signalling IL6 signalling unchanged
RA diseaseActivity lower
RA diseaseactivity higher
Protein target: IL6R
Biomarker Tocilizumab
IL-6 (n=1,446)
CRP (n=3,010)
Fibrinogen (n=409)Soluble IL-6R (n=1,465)
Albumin (n=108)
Haemoglobin (n=2,072)
Repurposing IL6R as a target for CHDThe Interleukin-6 Receptor Mendelian Randomisation Analysis(IL6R MR) Consortium* Lancet 2012; 379: 1214–24
People at risk of CHD
Random allocation of IL6R alleles
IL6R aa IL6R AA
IL6 signalling unchanged
CV eventrate lower
CV eventrate higher
Genetic study: natural randomisation
Reduced IL6 signalling
Protein target: IL6R
Drug interventionPatients with rheumatoid arthritis
Randomisation (Tocilizumab)
IL6R- blocker (MAB) Placebo
Reduced IL6 signalling IL6 signalling unchanged
RA diseaseActivity lower
RA diseaseactivity higher
Protein target: IL6R
Biomarker Tocilizumab
IL-6 (n=1,446)
CRP (n=3,010)
Fibrinogen (n=409)Soluble IL-6R (n=1,465)
Albumin (n=108)
Haemoglobin (n=2,072)
Repurposing IL6R as a target for CHD
Drug intervention
IL6R- blocker (MAB) Placebo
Reduce IL6 signalling IL6 signalling unchanged
RA diseaseActivity lower
RA diseaseactivity higher
Biomarker Tocilizumab IL6R SNP rs7529229
IL-6 (n=1,446) (n=29,838)
CRP (n=3,010) (n=76,527)
Fibrinogen (n=409) (n=52,667)Soluble IL-6R (n=1,465) (n=1,454)
Albumin (n=108) (n=5,787)
Haemoglobin (n=2,072) (n=17,898)
Patients with rheumatoid arthritis
Randomisation (Tocilizumab)
Protein target: IL6R
Genetic study: natural randomisation
Repurposing IL6R as a target for CHD
Additional examples• Darapladib, LpPLA2 and CHD
MR trial: Casas JP. et al. Circulation 2010 Jun 1;121(21):2284-93RCT: STABILITY N Engl J Med 2014 May 1;370(18):1702-11;SOLID TIMI 52 JAMA. 2014 Sep 10;312(10):1006-15
• Folic acid, homocysteine and stroke
MR trial: Holmes MV et al. Lancet 2011 Aug 13;378(9791):584-94RCT: Huo et al. JAMA 2015 Apr 7;313(13):1325-35
• Ezetimibe, LDL-C and CHD
MR trial: MI Genetics Consortium Investigators N Engl J Med 2014 Nov 27;371(22):2072-82RCT: Cannon CP et al. N Engl J Med 2015 Jun 18;372(25):2387-97
Published Genome-Wide Associations through12/2013
http://www.ebi.ac.uk/gwas/
GWAS ‘rediscoveries’ of human drug targetsGWAS Phenotype Associated Gene (Ensembl ID) Associated Gene Description Compound USAN/INN
Total/LDL cholesterolHMGCR(ENSG00000113161)
3-hydroxy-3-methylglutaryl-CoAreductase
Lovastatin,Pravastatin,Simvastatin
Type 2 diabetesKCNJ11(ENSG00000187486)
potassium inwardly-rectifyingchannel subfamily J member 11
Glyburide,Rosiglitazone
PPARG(ENSG00000132170)
peroxisome proliferator-activated receptor gamma
Rosiglitazone,Repaglinide
Nicotine dependenceCHRNA3(ENSG00000080644)
cholinergic receptor, nicotinic,alpha 3
Nicotine,Varenicline
CHRNB4(ENSG00000117971)
cholinergic receptor, nicotinic,beta 4
Nicotine,Varenicline
Courtesy Chris Finan and Felix Kruger
Illumina Human Drug Core – Array Design
Illumina Human Core Array
Whole genome tagSNP markers - 250,421Indel/exome markers >20,000Headroom for custom markers - 200,000
Drug development custom content
Illumina Human Drug Core
~480,179 assays and ~499,367 beadtypes
Targets of approved drugs and thosein clinical development; ADMET (~1426 genes)
Proteins with ‘drug-like compounds orclosely related to drug targets (~682 genes)
Extracellular or transmembrane targetsand members of drug target families(~2370 genes)
Variants of interest: GWAS SNPs; APOE; AIM;fingerprint
Developers: Casas, Finan, Shah, Kruger and Hingorani (UCL); Gaulton and Overington (EBI);together with the Illumina bioinformatics team
Fraction 1kg ph. 3 variantscovered (r2> 0.8)
10
Coverage of druggable genome by genotypingplatforms
Illu DrugDev Consortium 24
Tier 1 Tier 2 Tier 3a Tier 3b
CourtesyDr Chris Finan, UCL
Summary
• Genetic studies in populations share the design features of a randomised controlledtrial (RCT), the pivotal step in drug development
• Alleles in a gene encoding a drug target that affect its expression or activity canhelp predict the effect of modifying the same target pharmacologically
• Genetic studies in populations and patients may help support target selection andvalidation in drug development
Colleagues, collaborators and fundersPhilippa TalmudSteve HumphriesFotios DrenosSonia ShahDelilah Zabaneh
Harry HemingwayMartin BobakAida SanchezEric BrunnerMeena KumariMika KivimakiMichael Marmot
Mike HubankKerra PearceJutta PalmenDavid Balding
Chris PowerElina HyponnenJohn DeanfieldDi KuhAndy WongRichard MorrisPeter Whincup
Jacky Pallas
John WhittakerLiam SmeethFrank DudbridgeClaudio VerzilliLeonelo Bautista
Shah EbrahimDebbie LawlorTom GauntIan DayYoav Ben-ShlomoGeorge Davey SmithJackie PriceGerry Fowkes
Ann RumleyGordon LoweNaveed Sattar
Patsy MunroeToby JohnsonMark Caulfield
Manj SandhuClaudia LangenbergKen OngNick WarehamKay Tee KhawFrances WensleyJohn Danesh
Juan Pablo CasasMeena KumariTina ShahReecha SofatJorgen EngmannDan SwerdlowMichael Holmes
Rosetrees Trust
National Institute forHealth Research
Acknowledgements1. Institute of Cardiovascular Science, and
Farr Institute in London, UniversityCollege London, UK
2. Farr Institute in London, UniversityCollege London, UK
3. European Molecular Biology Laboratory -European Bioinformatics Institute,Cambridge, UK
4. Illumina Inc, San Diego, USA
5. Illumina UK Ltd, Little Chesterford, UK
– Chris Finan1
– Felix Kruger1
– Tina Shah1
– Jorgen Engmann1
– Juan-Pablo Casas1,2
– John Overington1,3
– Anna Gaulton3
– Anneli Karlsson3
– Rita Santos3
– Luana Galver McAuliffe4
– Ryan Kelley4
– Cora Vacher5
Acknowledgements
1. Institute of Cardiovascular Science, andFarr Institute in London, UniversityCollege London, UK
2. Farr Institute in London, UniversityCollege London, UK
3. European Molecular Biology Laboratory -European Bioinformatics Institute,Cambridge, UK
4. Illumina Inc, San Diego, USA
5. Illumina UK Ltd, Little Chesterford, UK
– Aroon Hingorani1
– Chris Finan1
– Felix Kruger1
– Tina Shah1
– Jorgen Engmann1
– Juan-Pablo Casas1,2
– John Overington1,3
– Anna Gaulton3
– Anneli Karlsson3
– Rita Santos3
– Luana Galver McAuliffe4
– Ryan Kelley4
– Cora Vacher5
Extending the use of genetic studiesto support target selection and validationin drug development
• The Druggable genome
• The design of a genotyping array to support target selectionand validation in drug development
The Druggable Genome
• With few exceptions, drug targets are proteins
• Not all proteins are amenable to targeting by the main classes oftherapeutics (small molecule drugs, therapeutic monoclonalantibodies or peptides)
• The ‘druggable genome’: defines the set of genes encoding druggabletargets
• ‘Druggability’ refers to the potential for a protein to be modified by adrug-like small molecule
Prior estimates of the Druggable Genome
• Predated contemporary estimates of the number of protein coding genes
• May not have considered targets of bio-therapeutic drugs (e.g. peptides andand therapeutic monoclonal antibodies)
• May not have included targets of recently licensed first-in-class drugs
An array with custom coverage of thedruggable genome
• Possibility that existing arrays either provided sparse coverageof druggable genes (e.g., GWAS arrays) or dense coverage of amodest number of druggable genes (e.g., gene-centric arrayssuch as metabochip, cardiochip etc)
• Advantage in having dense coverage of known and likely drugtargets across all disease areas
• Allow identification of tractable targets and drug repurposingopportunities
The Illumina Infinium DrugDev ArrayCo-developers: Casas, Finan, Shah, Kruger and Hingorani (UCL); Gaulton and Overington (EBI);together with the Illumina bioinformatics team
Potential users of the array
• Investigators with patient or population samples but noprior genotyping array
• Investigators with patient or population samples previouslygenotyped using a disease-focused fine-mapping array
• Investigators with patient or population samples genotypedusing earlier generation whole genome arrays
• Investigators contemplating genotyping of large-scaleelectronic health record datasets
Sivakumaran et al. Am J Hum Genet. 2011Nov 11; 89(5): 607–618
Pleiotropy in human complex diseases and traits
“……233 (16.9%) genes and 77 (4.6%) SNPs show pleiotropiceffects”
Disease_15 Disease_667 Disease_1123
Gene 1
Gene 20,000
Potential applications of the array
• Drug target discovery - identification of druggable proteins playing a causalrole in a disease of interest
• Drug target validation and prioritisation - informing if and when to advancean existing drug or drug-like compound through a drug-developmentpipeline
• Drug repurposing studies - identifying the role of a drug target in adifferent disease from the current drug indication
• Separating on- vs. off-target effects - for first-in-class and fast followerdrugs
• Stratified medicine studies - within randomised trials or non-randomisedresearch studies
Conclusions
• Genetic studies in populations, case collections and electronic health recorddatasets may help support drug development
• A new array which incorporates GWAS capability with custom content of thedruggable genome as well as genes involved in drug handlingmay help support such studies
• A consortium based on this new array is planned